Extending existing features with their non-linear mappings would improve logistic regression too, probably to the svc level (rbf or poly kernel is that but impl

Extending existing features with their non-linear mappings would improve logistic regression too, probably to the svc level (rbf or poly kernel is that but implicit). Linear models are really well researched, and today with the compute and with proper training and data preparation they can easily get to satisfying levels of performance for a variety of tasks.